The IoT dream behind SoftBank’s ARM purchase

For years, ARM has looked to be practically acquisition proof. Any company with a strong reason to buy the processor designer would be more than likely wreck ARM’s value proposition and with little upside.

For any company that wants access to ARM’s IP, there is little incentive to buy the entire company – all it takes is to become a partner with a licensing deal. It was possible to imagine an EDA company finally deciding to buy ARM but it would come at the risk of alienating customers who had bought into competing toolchains and with a punishingly high price tag. The slide in the value of the pound sterling relative to the dollar and the yen might have made ARM superficially less pricey – if you discounted the fact that, because its revenues are mostly earned in dollars, its share price over the past two weeks surged in the opposite direction to the pound.

In the end, the offer has come from an unexpected source: Japanese technology company SoftBank. At a price of more than $32bn, SoftBank appears to see value that goes way beyond ARM’s likely revenue streams. The company has grown rapidly since the last time speculation surged about the company’s independence – when the pound dropped sharply after the subprime debt crisis – practically trebling its sales to almost $1.5bn last year and seeing profits grow even more quickly. But the industry is looking at a situation where smartphone and tablet sales are approaching saturation at the high end and scraping away at margins on the low end.

Custom servers

ARM’s push into servers comes at a time when the server makers are increasingly focusing not just on heterogeneous computing using graphics processor units (GPUs) but programmable and custom logic. The larger server-farm operators are bringing more of the design in-house and have embarked on a program of making their largely open-source software base target independent. That’s not good news for Intel. It’s also not great news for ARM. The UK processor maker may find its customers displacing Intel Xeons in some cases, but with more of the silicon area being given over to custom SoCs perhaps built around open-source cores such as RISC V.

SoftBank’s stated aim is for ARM to continue to build up its role in the internet of things (IoT). However, there are issues here. The unit volumes are potentially massive but the sensor-node end of the IoT where Cortex-M cores are likely to dominate in the short term will place increasing pressure on licensing and royalty costs. Again, the open-source offerings could start to gain a foothold if SoftBank chooses to try to increase the value ARM obtains from future deals – and this is one of the Japanese company’s declared strategies.

Investment plan

In the short term, SoftBank aims to increase investment in ARM. It will, according to the offer documents, keep ARM’s HQ in Cambridge, England and it will double the staffing levels there, while increasing – by an unstated amount – staffing in ARM’s other locations. This may lead to an acceleration in IoT-oriented design work that will give the ARM architecture a stronger foothold in that market. Such an investment might prove effective at keeping customers invested in the architecture but in such as cost-sensitive market, the payback for SoftBank seems uncertain.

For ARM’s larger systems-house partners, the question for them is second-guessing SoftBank’s long-term strategy. As a revenue stream that places a massive premium and gamble on IoT growth, the acquisition does not look compelling. SoftBank is making changes on multiple fronts. It bought French robotics company Aldebaran as well as cellular operator Sprint. The overall thrust appears to be one of vertical integration around IoT and cyber-physical systems. The question becomes whether that vertical integration is enough to cause concern and convince them to start to pursue other options.